Polymer 44 (2003) 15771588

www.elsevier.com/locate/polymer

Probing the degree of crosslinking of a cellulose based superabsorbing

hydrogel through traditional and NMR techniques
F. Lenzia, A. Sanninoa, A. Borrielloa, F. Porroa, D. Capitanib, G. Mensitieria,*
b

a
Department of Materials and Production Engineering, University of Naples Federico II, P.le Tecchio 80, 80125 Naples, Italy
Institute of Chemical Methodologies, National Research Council, Research Area of Rome, via Salaria km 29.300, 00016, Monterotondo Stazione, Rome, Italy

Received 7 May 2002; received in revised form 19 December 2002; accepted 21 December 2002

Abstract
The network structure of a cellulose-based superabsorbing material has been probed by using three different techniques: 13C solid state
NMR, free swelling in water and uniaxial compression of water swollen samples. A good agreement between the three apporaches has been
found in terms of concentration of crosslinks per unit volume.
The results have been discussed taking into account that NMR technique is able to detect only chemically effective crosslinks while free
swelling and compression are sensitive to elastically effective physical and chemical crosslinks.
A depression of swelling capacity and an apparent increase of degree of crosslinking with time, promoted by ageing of the cellulosic
material, has been experimentally evidenced and discussed in terms of development of intermolecular physical interactions.
q 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Degree of crosslinking; Cellulose; Swollen hydrogel

1. Introduction
Superabsorbing polymeric materials have a wide range
of applications, mainly in the field of personal hygiene
products [1,2], which represents 80% of the hydrogels
production nowadays, with more than 8000 tons demand in
1999 only on the european market. An important focus of
the research in this field is materials biodegradability.
Modern superabsorbents are acrylamide based products and,
as such, non-biodegradable. The renewed attention of
Institutions and public opinion towards environmental
protection issues sensibilized some producers in the
development of biodegradable superabsorbents. Potentially
biodegradable cellulose based superabsorbent can be
synthesized with sorption properties similar to those
displayed by acrylate based products [3,4]. These materials
can be obtained by chemical crosslinking of cellulose
polyelectrolyte derivatives using small difunctional molecules as crosslinkers, which creates intermolecular
covalent bonds among polymer molecules forming a
three-dimensional hydrophilic network. Further improve* Corresponding author. Tel.: 39-817682512; fax: 39-817682404.
E-mail address: mensitie@unina.it (G. Mensitieri).

ment of water sorption capacity can be obtained by inducing

the formation of a microporous structure [4].
Swelling properties of crosslinked hydrophilic polyelectrolites in aqueous solutions are determined by several
chemical and structural factors, i.e. hydrophilicity of
polymer backbone, degree of crosslinking, density and
type of fixed charges, degree of ionization and amount of
ionic groups and eventual presence of porosity. The relative
importance of these factors depends upon the properties of
aqueous solution in contact with the network, i.e. pH, ionic
strength, presence of other solvents.
The free energy change associated with the swelling
process is made of several contributions [5 11] that are
related to polymer/solvent mixing, elastic response of the
network, presence of bound and mobile ions and electrostatic interactions among fixed charges present on the
polymer chains. These factors, in turn, are related to the
above mentioned chemical and structural factors.
In particular, the effect of entropic elasticity of the
network, which becomes of major importance at high
swelling degree, is strongly dependent upon the degree of
cross-linking. In fact, water sorption capacity of crosslinked
polymeric networks is affected significantly by the elastically effective degree of crosslinking, which takes into

0032-3861/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 3 2 - 3 8 6 1 ( 0 2 ) 0 0 9 3 9 - 4

1578

F. Lenzi et al. / Polymer 44 (2003) 15771588

account the effect of all segments of macromolecular chains

which react elastically to their elongation promoted by
network swelling. Crosslinks, which act as constraints to the
movement of segment ends anchoring them to the
network, can be both chemical and physical: the elastically
effective chemical crosslinks are all binding points among
chains with the exclusion of those of them generating
dangling ends and loops. The elastically effective
physical crosslinks are, for example, chain entanglements
[12] or strong intermolecular interactions. Conversely, the
chemically effective crosslinks are intended as all the
chemical crosslinks, both elastically effective and not.
The degree of crosslinking (d.c.) is defined as the number
density of junctions joining the chains into a permanent
structure. According to this definition the degree of
crosslinking is given by [5]
d:c:

n
2V

where n/2 is the total number of chemical cross-links and V

is the total volume of polymer.
In the case of a perfect network with no dangling ends,
loops and entanglements, which could be obtained by
joining pair of segments of linear chains through chemical
cross-links, the concentration of elastically effective chain
elements, rx ne/V, corresponds to the concentration of all
chemically crosslinked polymer segments (n/V):

rx

ne
n
1


v MC
V
V

 C is the
where v is the specific volume of the polymer, M
average molecular weight between cross-links, and ne/V are
the moles of elastically effective chains per unit volume of
network. According to the definition given above, rx and
rx/2 represent, respectively, the moles of polymer segments
engaged by crosslinks and the moles of crosslinks per unit
volume of the network.
Generally, the determination of elastically effective
crosslinking density can be performed, among other
techniques, by using equilibrium swelling measurements
or by performing uniaxial compression tests on swollen
networks [5,13 16]. These techniques allow a straightforward determination of crosslinking density if the chemical
structure is simple and structural complexities (e.g. fixed
ionic charges) are absent. In fact, for crosslinked polyelectrolites systems (on which superabsorbing materials are
mostly based) the determination through swelling measurements can be performed only if it is possible to evaluate
quantitatively the contributions arising from the presence of
charges and degree of ionization in the actual experimental
conditions [6 11]. A comprehensive collection of equilibrium swelling models in aqueous solutions for chemically
crosslinked ionic gels is reported in Ref. [6], where the
different contributions of the involved mechanisms are
discussed. In turn, if the crosslinking density is determined
through mechanical testing (uniaxial compression test) on

equilibrium swollen gels [11 16], simplified approaches

are often adopted to interpret experimental data, assuming,
in general, that the interactions between charges does not
affect the mechanical response and that the volume of the
gel does not change during the test.
Both experimental approaches supply an overall value
for the degree of crosslinking (chemical physical crosslinking) and rely upon theoretical models for description of
polymer/solvent mixing, network elastic behavior and effect
of fixed charges present on polymer backbone.
The availability of an alternative experimental technique
which is able to supply only the value of the chemically
effective crosslinking density, could be an important aid for
the characterization of crosslinked systems, both in the field
of hydrophilic networks forming highly swollen hydrogels,
and of thermosetting polymeric systems in general. To this
aim, Solid State NMR spectroscopy repesents a nondestructive approach to gather information on the chemical
structure of these amorphous, insoluble materials [17 34]
independently of the knowledge of other chemical physical
and structural parameters (such as density of charges,
degree of ionization and polymer/solvent interaction
parameter). This approach has the advantage of not being
affected by the eventual sample microporosity and that no
assumptions have to be made or theoretical models have to
be adopted to describe the behavior of the network. The
main limit is that it accounts also for non-elastically
effective chemical crosslinks (those generating loops or
dangling ends) while not accounting for physical crosslinks.
In this contribution we compare the results obtained by
using three different techniques, i.e. free swelling measurements, uniaxial compression and 13C solid state NMR, to
probe the degree of crosslinking of a cellulose based
superabsorbing hydrogel synthesized by crosslinking with
divinylsulphone a mixture of hydroxyethylcellulose and
sodium salt of carboxymethylcellulose, in an aqueous
medium.

2. Theoretical background
2.1. Determination of elastically effective degree of
crosslinking from free swelling equilibrium
Swelling equilibrium of a polymer with a solvent is
obtained when the chemical potentials of the solvent in the
polymer phase and in the free solution are the same. This
statement can be expressed in terms of osmotic swelling
pressure as:

PMIX PELAS PION PELEC 0

This relationship is based on the assumption that the

contributions are simply additive. The first term takes into
account the mixing tendency of polymer and solvent. The
second accounts for elastic response of the network due to
the constrain imposed by the presence of chemical and

F. Lenzi et al. / Polymer 44 (2003) 15771588

physical crosslinks. The third and fourth terms, which have

to be included if ionizable groups are present, are due,
respectively, to differences in mobile ions concentration
between gel phase and external solution and to the
electrostatic interactions of charges on the polymer backbone. This last term is generally negligible in comparison to
the third one in the case of highly swollen systems.
Several theories can be adopted to express mathematically the three prevailing contributions (i.e. PMIX, PELAS
and PION). In the present contribution, Flory Huggins
theory [5] has been adopted for the mixing term. For the
evaluation of the elastic term the hypotheses [5] of affine
deformation and of Gaussian distribution of molecular
chains between two crosslink points have been adopted, and
the presence of solvent during crosslinking reaction has
been accounted for (in fact polymer volume fraction in the
starting reacting solution, f2,r, is smaller than 1). For the
evaluation of the ionic term, it has been considered the case
of an anionic polymer nework in which the ionic
concentration in the external solution is negligible if
compared to the ionic concentration inside the gel (ionic
strength of the external solution 0). This hypothesis is
appropriate in the case at hand, in view of the moderate level
of crosslinking of the network and of the circumstance that
only the equilibrium of the gel with distilled water has been
analysed.
A thorough treatment, reporting the analysis of a wider
range of cases (e.g. non-Gaussian statistics, high ionic
strength of the swelling solution) can be found in Ref. [6].
For the definition of thermodynamic equilibrium, Eq. (3)
should be coupled to the expression for the equilibrium
between ionic species in the gel phase and in the external
solution; as first approximation, we imposed that concentrations are the same inside and outside the gel [5].
On the basis of the assumptions discussed above, the
following expression for rx can be derived [6,7]:
2

rx

ln1 2 f2 f2 xf2 
if2

v1
z vm
(
!1=3
!)
f2
f2
f2;r
20:5
f2;r
f2;r

where vm is the molar volume of structural repeating unit of

the network; rx, the moles of polymer units engaged by
crossinks per cm3 of dry network; f2,r, the polymer volume
fraction in the reaction mixture before full swelling in
distilled water (polymer volume fraction in the relaxed
network); v1, the molar volume of swelling solvent; i, the
fraction of polymer structural units which carry ionized
couples; x, the Flory Huggins interaction parameter
(polymer/swelling solvent); f2, the polymer volume fraction at swelling equilibrium and z is the number of
electronic charges carried by cations formed through
dissociation of ionic groups present on polymer backbone.
In Eq. (4): 2ln1 2 f2 f2 xf2 =v1 is the polymer/
water mixing term. if2 =z vm is the contribution related to

1579

the presence of ionized ionic couples. f2;r {f2 =f2;r 1=3 2

0:5f2 =f2;r } is the entropic elastic contribution related to
the retractive tendency of macromolecular segments; it
takes into account that reaction has been performed at
f2,r , 1.
2.2. Detemination of elastically effective degree of
crosslinking from uniaxial compression of swollen networks
A swollen hydrogel, when submitted to a uniaxial
compressive load, displayes a deformational behaviour
which depends upon the elastic response of deformed
chains, to the interaction among fixed charges and to the free
energy change associated to the release of some amount of
sorbed water. By making the simplifying assumption that no
volume change occurs upon compression of the swollen
hydrogel, Flory [5] derived a relationship between the
compressive stress and the compressive deformation for
the case of a swollen croslinked polymer, based on the
assumption of Gaussian statistics and of affine deformation.
This expression can be easily extended to the case of
f2,r , 1 obtaining [35]:

s RT





ne 2=3 1=3
1
1
f2;r f2;s a 2 2 G a 2 2
V0
a
a

where s F/A0 is the uniaxial compressive stress (where F

is the traction force and A0 is the initial area of swollen
sample cross-section), a L=Li ; with L, the actual thickness
of the compressed swollen sample and Li, the initial
thickness of the swollen sample, R, the universal gas
constant, T, the absolute temperature, f2,s, the polymer
volume fraction in the swollen state under compression
which is assumed to be equal to the value for the
undeformed swollen gel, ne/V0, the moles of elastically
effective chains per cm3 of dry polymer network and G is
the shear modulus of the swollen network.
Although this approach is oversimplified, due to the
assumption of constant volume (actually some water is
squeezed out of the swollen sample as a result of
compression), it is suitable to interpret gel deformation
behaviour under uniaxial compression in the case of small
deformations (a ! 1) and can be used for the evaluation of
the ratio ne/V0 and of rx.
The deviation from Gaussian statistics at large deformations can be taken into account by using a phenomenological expression to describe the behaviour of a swollen
crosslinked network submitted to uniaxial extension. This
expression can be derived from the expression of Mooney
Rivlin strain energy function [36] for swollen rubbers which
accounts for both the deformation due to swelling and the
deformation due to compression. Making the assumption of
incompressibility, the following expression relating the
uniaxial stress s (referred to the cross-sectional area of
undeformed swollen sample) to the extension ratio, a, can

1580

F. Lenzi et al. / Polymer 44 (2003) 15771588

Scheme 1. Schematic structure of CMCNa-HEC crosslinked with DVS.

be derived:



K2
1
s 2 K1
a2 2
a
a

3. Experimental
3.1. Materials
6

where s has the same definition as in Eq. (5) and the values
of K1 and K2 are proportional to the swelling ratio of the
sample. According to Eq. (6), a plot of sa 2 1=a2 21 vs.
1/a based on experimental data should be linear.
2.3. Detemination of chemically effective degree of
crosslinking from solid state 13C NMR measurements
13

C Solid State NMR analysis can be used to determine

the chemically effective crosslinking degree of a polymer
network. The application of this experimental procedure to
the system under investigation is reported in full details in
Ref. [37]. In brief, the analysis can be carried out by
performing 13C cross polarization magic angle spinning
(CP-MAS) solid state spectra and single pulse excitation
(SPE) experiments. It consists in evaluating the moles of
crosslinks per polysaccharide ring by the ratio of half of the
area of the resonance peak associated to methylene carbon
atoms present on reacted crosslinker molecule (C atoms
labeled A in Scheme 1) to the area of resonances due to
anomeric carbon atoms, C1, present on polysaccharide rings
(labeled 1 in Scheme 1). By using this procedure to
evaluate the chemically effective degree of crosslinking, we
have implicitly neglected reactions occuring between DVS
and water and DVS oligomerization. The eventual occurrence of these reactions, which could cause errors in the
NMR calculations, could be assessed by proper procedures
of hydrolisis of crosslinked polymer and subsequent
quantitative analysis of resulting fractions.

Highly absorbing cellulose based hydrogels synthesized

in this work have been obtained by crosslinking water
solutions of carboxymethylecellulose sodium salt (referred
in the following as CMCNa) and hydroxyethylcellulose
(referred in the following as HEC) using divinylsulphone as
crosslinking agent (referred in the following as DVS).
CMCNa (cod. 41,933-8), HEC (cod. 30,683-3) and DVS
(cod.V370-0) were purchased from Aldrich Chimica s.r.l.
Milano and used as received. The degree of substitution
(D.S.) of CMCNa was 0.66 ^ 0.03 [38], while the D.S. for
HEC was about 1. Further details on starting materials are
reported in Refs. [37,38].
Crosslinking reaction has been performed according to
procedures reported in the literature [39]. For this study a
mixture of CMCNa and HEC, with a weight ratio equal to
3/1, was first dissolved in a solution of DVS and distilled
water by stirring gently at room temperature until a clear
solution was obtained. The total polymer weight fraction in
the solution was 2%. Two different DVS concentrations
were used in order to obtain samples with two degrees of
crosslinking: 0.04 and 0.133 moles of DVS per liter of
solution. In the following, crosslinked samples prepared
with a starting DVS concentration equal to 0.04 and
0.133 mol/l will be referred, respectively, as sample L
(low crosslinking density) and sample H (high crosslinking density). The presence of HEC is necessary to
promote quantitatively intermolecular rather than intramolecular crosslinking [39]. After the mixing stage, an aqueous
solution of KOH was added as catalyst to the polymer
solution until a pH 12.5 was reached. Hydrogel formation
occurred in few hours at room temperature. A schematic

F. Lenzi et al. / Polymer 44 (2003) 15771588

representation of cross-linked chemical structure is reported

in Scheme 1.
To remove unreacted DVS and the KOH, the partially
swollen hydrogel was cut in small pieces which were then
soaked in distilled water reaching equilibrium swelling
under continuous stirring (washing stage). Water was used
in volumes greatly exceeding that of hydrogel and was
changed several times. These equilibrium swelling data,
obtained both for 0.04 and 0.133 moles of DVS per liter of
starting solution, have been used (see Section 4) for the
determination of elastically effective crosslinking degree
before desiccation took place. Typical values of free
swelling right after the synthesis were around 880 and
140 g of water per gram of dry polymer, respectively, in the
case of sample L and H.
After the washing stage, a dry product was obtained by
desiccating hydrogel samples by phase inversion with
acetone according to procedures reported in literature [40]
and then kept under vacuum for some hours in order to
remove residual acetone. This desiccation procedure was
used since the high and connected microporosity induced by
the fast water extraction procedure guarantees very high
water swelling capacity, which was one of the objectives of
the present research. The samples obtained with this
procedure were in form of a coarse powder. A sample was
considered to be desiccated when the water content was
lower than 10% by weight of dry polymer. For the sake of
comparison, uncrosslinked samples were also prepared
(referred in the following as sample U). For this purpose
the starting aqueous solution of CMCNa, HEC and DVS
(total polymer concentration 2% by weight, CMCNa/HEC
ratio equal to 3/1 and DVS concentration equal to
0.133 mol/l) was put directly in contact with acetone and
the precipitate collected. Both types of samples were then
grinded under liquid nitrogen to obtain a fine powder to be
used for NMR solid state analyses.
To perform reliable compressive tests, samples with the
lower degree of crosslinking (0.04 moles of DVS per liter of
starting solution) were also obtained in the form of thin films
(around 50 mm thickness). The procedure was identical to
the one reported above till the initial stage of the reaction. In
fact, few seconds after the addition of catalyst and the
mixing of the components, the reacting solution was poured
on teflon trays to be desiccated under a mild convective air
flow at 50% relative humidity. Regular films were thus
obtained after a 48 h desiccation. Also in this case, samples
were considered to be desiccated when the water content
was lower than 10% by weight.
Wide angle X-ray scattering analysis performed on
crosslinked samples indicated that the samples were totally
amorphous [40].
3.2. Methods
3.2.1. Solid State NMR measurements
Finely powdered samples H, L and U were packed into

1581

4 mm zirconia rotors and sealed with Kel-F caps. Solid state

13
C CP-MAS NMR spectra [41] were run on a Bruker AC200 spectrometer, equipped with an HP amplifier for 1H at
200 MHz, 120 W CW and with a pulse amplifier M3205.
The spin rate was always kept at 8 KHz. The p/2 pulse
width was 3.1 ms and the recycle delay was 4 s. Spectra
were obtained with 1024 data points in the time domain,
zero-filled and Fourier-transformed with a size of 2048 data
points; 19,000 scans were performed for each experiment.
As a function of the contact time, two series of
experiment were performed, with the contact time ranging
from 0.02 to 7 ms.
Magic angle spinning with single pulse excitation (MASSPE) experiments were performed, the recycle delay was
40 s, the 13C p/2 pulse width was 2.8 ms. With this
experiment a single p/2 pulse is applied to excite the
carbon spectrum which is recorded in the presence of MAS
and dipolar decoupling (DD) [42].
Analysis of NMR resonances was performed using the
deconvolution program GLINFIT [43]. This program can
perform the full deconvolution of overlapped lines both with
Gaussian and/or Lorentzian shapes. Line widths, chemical
shifts and areas along with their standard deviations were
obtained. In the case reported here, the standard deviations
on the obtained areas were about 10% of their nominal
value.
All tests were performed at 25 8C.
3.2.2. Uniaxial compression tests
Swollen films were placed between two flat plates made
of polymethylmethacrylate. Both plates and sample were
immersed in a temperature controlled water jacketed
container filled with distilled water to prevent drying of
swollen network. The upper plate was contacted with a load
cell (Burster type 8453, max load 1000 N, sensitivity 0.1 N)
attached to a movable frame. The frame carrying the load
cell can be moved in extremely small steps monitored by a
travelling microscope (accuracy ^ 2 mm). Each measuring
step consisted in imposing a constant strain and in
measuring the associated compressive stress. Once a
constant stress value was obtained, it was taken as the
equilibrium value and strain was, then, increased further.
The design of the compression apparatus allows a
continuous contact of the film with distilled water and the
cone and cup geometry prevents the relative slipping
between upper and lower plate. All tests were performed
at 25 8C. For each type of sample, experiments were
repeated four times and average values of stress as function
of imposed deformation are reported.
3.2.3. Equilibrium free swelling tests
Equilibrium free swelling was measured by first
equilibrating in distilled water small pieces of gel, right
after the washing stage and then weighing samples removed
from the water using a Mettler AE 100 electronic balance
with a sensitivity of 1025 g. Before measuring the weight,

1582

F. Lenzi et al. / Polymer 44 (2003) 15771588

samples were quickly and gently blotted to remove liquid

water from the surface and placed into sealable plastic
containers of known weight. Samples were then desiccated
at 50 8C under high vacuum (1023 Torr) to determine the
dry polymer weight. Equilibrium swelling was then
expressed as ratio between grams of absorbed water and
grams of dry polymer. All tests were performed at 25 8C.
3.2.4. Water vapour sorption isotherms
Water vapour sorption isotherms were determined at
25 8C by hanging the powdered sample, placed in a glass
pan, to a quartz spring microbalance (Ruska Co., Tx, USA,
full elongation 400 mm, max. weight 50 mg) operating in a
water vapour environment at accurately controlled temperature and pressure. The spring was placed in a water
jacketed chamber with service lines to a reservoir of
degassed liquid water, to an accurate pressure transducer
(M.K.S. Baratron 121 A, with a full scale of 100 Torr and a
sensitivity of 0.01 Torr) and to a turbomolecular pump.
Before each sorption run, samples were desiccated under
vacuum. Water vapour was then admitted in the cell at the
selected pressure. Weight increase due to water sorption
was monitored by measuring the spring elongation with a
travelling microscope (sensitivity of 0.01 mm) until a
constant equilibrium weight was reached. Equilibrium
sorption values were then reported vs water vapour activity
as gram of water absorbed per gram of dry polymer. The
water vapour activity at which each sorption run was
performed has been evaluated as the ratio p/p0, where p is
the pressure of water vapour at which was performed the
sorption run while p0 is the water vapour pressure at the test
temperature.

4. Results and discussion

In the following, we present separately the results
obtained by using the three experimental approaches
which will be reported as moles of effective chains per
cm3 of dry crosslinked network (rx). It is worth noting that
some terms used in the determination of degree of
crosslinking from swelling measurements were obtained
from NMR analysis, as described in the following sections.
As a consequence the experimental errors in the evaluation
of rx from swelling measurements, is due both to the errors
related to the evaluation of swelling ratio and to the errors
related to the determination of some terms from NMR
measurements.
Another important point of concern is that the crosslinked polysaccharide based systems under investigation are
expected to slowly and progressively develop intermolecular interactions during and also after desiccation, as
witnessed by a decrease of swelling capacity of the sample
with the time of storage in dry conditions. As a
consequence, the elastically effective degree of crosslinking
(chemical physical) increases with ageing of network,

likely due to the development of intermolecular interactions, which act as additive physical crosslinks and
promote a depression of swelling capacity. Also the effect of
the progressive chain association on NMR measurements
cannot be ruled out. For this reason the results of swelling,
compression and NMR experiments used for the comparison of different techniques, refer only to samples right after
their synthesis.
4.1. Degree of crosslinking from free swelling
measurements
Swelling tests were performed in distilled and deionized
water right after reaction and washing stage, measuring a
swelling degree equal to 880 ^ 50 g/g and 140 ^ 10 g/g,
respectively, for sample L and H. Eq. (4), was used to
evaluate rx from these data. To perform this calculation,
some network parameters must be available: molar volume
of repeating unit of the network (vm), degree of ionization
(i), Flory Huggins interaction parameter (x).
The value of vm, molar volume of equivalent repeating
unit of the network was evaluated from relative amounts of
components as measured by NMR: (see Section 3.2.1)
contents of reacted DVS equal, respectively, to 0.04 ^ 0.01
and 0.33 ^ 0.04 moles per mole of polysaccharide ring
were found, respectively, in the case of sample L and
H. These results indicate that around 9 and 21% of DVS
present in the initial reacting solution was actually involved
in the crosslinking reaction, respectively, for L and H
sample. The resulting average molecular weights of the
equivalent repeating units were 218.4 for sample L and 200
for sample H. Assuming for the dry polymer network a
density equal to 1 g/cm3, the value of vm is numerically
equal to the determined values of the average molecular
weight of the equivalent repeating units. As a consequence,
the values of 218.4 and 200 cm3/mol were, respectively,
assumed in the case of sample L and of sample H.
The value of i, degree of ionization, can be evaluated
once the degree of substitution (DS) of CMC is determined
through high field NMR experiments or by conductimetric
titration. We assumed, for the interpretation of swelling in
distilled water, that i iMAX since all ionic groups are
likely to be ionized. This assumption is supported by the fact
that swelling was measured at pH 7, which is a condition
promoting rather complete ionization of the carboanionic
COO2 groups on the polymer backbone. Since i expresses
the fraction of repeating units carrying ionized ionic
couples, an equivalent repeating unit is introduced in
order to calculate the degree of ionization: it is a molar
average of CMCNa and HEC repeating units and DVS
molecular weights as determined on the basis of NMR
evaluation of their relative amounts. From the D.S. value of
CMCNa (0.665 ^ 0.03, see Ref. [38]) it can be easily
inferred that iMAX of pure CMCNa (for the case of pH . 7)
is 0.665. Consequently, on the basis of the average
molecular weight of the equivalent repeating unit of the

F. Lenzi et al. / Polymer 44 (2003) 15771588

1583

Table 1
Values of relevant terms appearing in Eq. (4) and numerical values with standard deviations of the mixing, elastic and ionic contributions for samples L and H

vm
v1
x
f2,r
f2
2ln1 2 f2 f2 xf2 =v1
if2 =vm
f2;r {f2 =f2;r 1=3 2 0:5f2 =f2;r }
a

Sample L

Sample H

218.4 cm3/mola
18 cm3/mol
0.98
0.0204
1.14 1023 ^ 0.05 1023
23.46 1028 ^ 0.3 1028 mol/cm3
2.6 1026 ^ 0.1 1026 mol/cm3
7.23 1023 ^ 0.1 1023 mol/cm3

195 ^ 5 cm3/mola
18 cm3/mol
0.98
0.0235 ^ 0.001
7.14 1023 ^ 0.5 1023
21.35 1026 ^ 0.2 1026 mol/cm3
1.45 1025 ^ 0.2 1025 mol/cm3
1.20 1022 ^ 0.05 1022 mol/cm3

Assuming density of bulk polymer 1 g/cm3.

crosslinked structure, values of iMAX for samples L and H

were calculated to be, respectively, 0.500 ^ 0.020 (about
50% of repeating unit carry an ionized group) and
0.390 ^ 0.020 (about 40% of repeating unit carry an
ionized group).
Flory Huggins interaction parameter has been evaluated
by fitting water vapor sorption isotherm for crosslinked
network at 25 8C using the model introduced by Flory [5] to
evaluate the chemical potential of a solute dissolved in a
crosslinked polymer, as modified to take into account that
crosslinking reaction has been performed at f2,r , 1:
ln a1

m1 2 m01
RT

ln1 2 f2 f2 xf22 
|{z}
b

v1 ve
1=3 2=3
v0 f2 f2;r 2 0:5f2
|
{z}

where v1 is water molar volume, a1 is water activity, m1 is

the chemical potential of water in the mixture, m01 is the
chemical potential of pure liquid water. For equilibrium
sorption with a water vapor at an activity a1 in the range 0
0.8, a term is , 0 since f2,r , 1. This implies that sorbed

amount for a crosslinked polymeric system should be higher

than that for the same uncrosslinked polymer system.
Actually the experimental results (see Fig. 1) points to the
opposite, thus indicating that in this activity range the
contribution of term a is much lower than contribution of b
term. In other words, the change of x due to modifications in
the chemical composition of polymeric system promoted by
crosslinking reaction, overwhelms the effect associated to
the entropic responce of macromolecular segments which, if
any, would increase, instead, sorption capacity. In fact, it
can be readily seen that in the range a 0.2 0.6 the
numerical value of a term, in the right hand side of Eq. (6),
is less than 1% of the value of b term. As a consequence
data fitting has been performed in the range 0 0.6
neglecting the term a and using only term b.
Water vapor sorption isotherm was evaluated for samples
U and L immediately after desiccation, before the
interactions related to network aging develop. The values
of x as determined by using Eq. (7) were 0.77 and 0.98,
respectively, for sample U and L.
The difference of the values for x are related to the
difference in the chemical structure of sample L as
compared to sample U, which is characterized by a lower
hydrophilicity due to the presence of reacted DVS
molecules. It is worth noting that the parameter x is a
function of polymer volume fraction expressed, in general,
by a Taylor series expansion [44,45]:

x x1 x2 f2 x3 f22

Fig. 1. Water vapour sorption isotherms at 25 8C for powdered samples U

and L. Lines are drawn to guide the eye.

where x1, x2 and x3 are independent of concentration. For

the evaluation of mixing contribution in Eq. (4) we adopted
a constant value for x, equal to 0.98. This assumption is
justified by the fact that in the case of highly swollen, fully
ionized networks, the effect of changes in parameter x on
swelling behaviour is rather negligible (see for example Ref.
[6]).
In Table 1 are summarized the calculated values for the
relevant terms appearing in Eq. (4) along with the numerical
values of the mixing, elastic and ionic contributions for both
samples.
Using Eq. (4), rx values were then determined
obtaining 3.54 1024 ^ 0.25 1024 and 10.60 1024
^ 2.5 1024 moles of crosslinks per cm3 of crosslinked

1584

F. Lenzi et al. / Polymer 44 (2003) 15771588

 C values,
dry polymer, respectively, for sample L and H. M
as evaluated from Eq. (2), were, respectively, equal to about
900 and 3000 g/mol, which are rather small values. For this
reason, the degree of crosslinking was also evaluated by
using an alternative expression for the elastic term
accounting for non-Gaussian chain statistics, which is
 C value is small [6,46]. Since
more appropriate when M
results obtained were only slightly different from those
based on the hypothesis of Gaussian statistics (a value for rx
was obtained which was about 15% smaller), this assumption was considered adequate for the sake of comparison of
rx as evaluated from swelling experiments with the results
of the other techniques.
It is worth of note that, as anticipated previously,
equilibrium swelling properties of the investigated hydrogels
in distilled water have been found to steadily decrease with
the time elapsed after desiccation stage and before swelling
test. This behavior is not related to a further decrease of
absorbed water in the dried polymer, which has been found to
remain constant, through thermogravimetric analysis. It is
likely that during the time interval between desiccation and
re-swelling of the films, physical interactions developed
among polysaccharide chains involving electron donor
acceptor groups. These interactions imposed a further
mobility constraint to macromolecules, acting as additive
physical crosslinks. Since these interactions are expected to
increase with time, sample aged for longer times consistently
displayied a lower sorption capacity related to the increase of
entropic response of the network.
4.2. Degree of crosslinking from uniaxial compression tests
The shear modulus of the swollen gel, G, to be used to
determine the degree of crosslinking from uniaxial compresion tests, was evaluated in the limit of small deformations
(a 1). In particular, the numerical value has been obtained
from a MooneyRivlin plot [36] of compression data (see
Fig. 2) by extrapoliting to the Y axis a linear fitting of data. This

Fig. 2. Mooney-Rivlin plots for equilibrium swollen films of type L,

evaluated at 25 8C: lines represent linear fittings.

analysis was performed only on sample films L. In Fig. 3

experimental results are reported as compressive stress (s) vs.
a, along with those predicted adopting in Eq. (5) the value of G
determined as discussed above. Data fitting is quite acceptable
down to a value of a 0.35: for lower values, the simplified
approach on which Eq. (5) is based, is inadequate.
The value of rx obtained from these measurements for
sample L was 2.4 1024 ^ 0.3 1024 moles of crosslinks
per cm3 of crosslinked dry polymer, which is in close
agreement with the value obtained from swelling
experiments.
Compressive measurements performed on aged sample,
clearly show an increase of apparent degree of crosslinking
with the time of aging as can be inferred from Figs. 2 and 3
(curves at 7, 14 and 21 days after desiccation). Similar
trends were obtained from swelling measurements. Values
of apparent crossliking degree as function of aging time, as
determined from swelling and compression tests, are
reported in Fig. 4. The reason for the mismatch, at long
desiccation times, between the values of rx as evaluated by
the two techniques has not yet been elucidated.
4.3. Degree of crosslinking from NMR measurements
4.3.1. 13C CP-MAS NMR spectra
In Fig. 5(a) the 13C CP-MAS spectrum of a physical
mixture obtained by mechanical mixing of CMC-Na and
HEC powders (3/1), is shown.
At 104.0 ppm the resonance of anomeric carbons C1 of
anyhydro D -glucose units unsubstituted in the C2 position is
observed, while at 97.8 ppm the signal of the anomeric
carbon C1(2p) of units substituted in position 2 is observed.
The shoulder at 82.7 ppm is mostly due to C4.
The intense and broad resonance centred at about
75.1 ppm is mostly due to C2, C3 and C5, however also
resonances due to the oxyethylene chains of substituents

Fig. 3. Uniaxial applied compressive stress as function of a, for equilibrum

swollen films of type L, measured at 25 8C. Lines represent the values
predicted through Eq. (5): arrows mark the values of a at which deviations
from experimental results become remarkable.

F. Lenzi et al. / Polymer 44 (2003) 15771588

1585

56.0 ppm is due to methylene carbons of type A (see

Scheme 1) i.e. to carbons adjacent to the sulfoxide group,
while methylene carbons of type B resonate in the same
crowded range of frequency of C2, C3 and C5, and cannot
be resolved in a solid state spectrum. It is worth of note that
the resonance at 56.0 ppm is fully absent in the spectrum of
the uncrosslinked sample (see Fig. 5(a)).

Fig. 4. Time evolution of rx for films of type L, as evaluated from swelling

and from uniaxial compression tests. Lines are drawn to guide the eye.

resonate in this range of frequency. At 62.7 ppm the

resonance of unsubstituted C6 carbon atoms is observed.
In Fig. 5(b) and c the spectra of sample L and the
spectrum of sample H are reported. The resonance at

4.3.2. Semi-quantitative evaluation of carbon spectra at the

solid state
Since the intensity of carbon resonances depends on the
cross-polarization rates, 13C CP-MAS spectra are not
quantitative.
The question rises when semi-quantitative information
must be obtained. To this aim, the cross-polarization
dynamic has to be carefully investigated [47].
The cross-polarization dynamics can be described by the
equation:
!



S0
2lt
2t
9
St
1 2 exp
exp
TIS
l
T1r 1 H

l1

TIS
TIS
2
13
T1r C
T1r 1 H

where S0 is the area and/or intensity of the investigated

resonance at t 0, T1r(1H) and T1r(13C) are the proton and
the carbon spin-lattice relaxation times in the rotating frame
and TIS is the cross relaxation time between protons and
carbons.
In homogeneous systems in which T1r(1H) is singlevalued, i.e. all resonances show the same T1r(1H) value
within the experimental error, the S0 values are the true
intensities and/or areas of the resonances under investigation. Under these circumstances signals will be in the
correct intensity ratio allowing a semi-quantitative analysis
[48].
A different approach must be used when carbon
resonances show different T1r(1H) values. In such a case
the MAS-SPE experiment is mandatory for obtaining a
semi-quantitative analysis. However this experiment has
two main drawbacks: since no cross-polarization from
protons to carbons is applied, the signal/noise ratio is poor;
the recycle delay (RD) between successive scans must be
long enough to allow all carbon sites to relax back to
equilibrium. The recycle delay strictly depends on T1 (13C)
values. In many rigid glassy or crystalline polymers T1 (13C)
may be very long, up to 102 104 s. Note that for a
quantitative measurement: RD . 5 T1 (13C). Hence in many
cases this approach is unrealistic.

Fig. 5. 13C CP-MAS NMR spectra of (a) a physical mixture of an

uncrosslinked CMC-Na /HEC powder (3/1); (b) sample L with low degree
of crosslink; (c) sample H with high degree of crosslink. The assignment is
also reported.

4.3.3. Degree of crosslinking in CMCNa/HEC/DVS

networks
13
C CP-MAS spectra of sample H have been run as a
function of the contact time t, with t ranging from 0.02 to
7 ms. The area of few selected carbon resonances has been
reported against the contact time (see Fig. 6). Fitting the

1586

F. Lenzi et al. / Polymer 44 (2003) 15771588

Fig. 6. Correlation between the areas of resonances of the 13C CP-MAS

spectrum of sample H and the contact time t. Lines through experimental
points are obtained fitting the experimental data to Eq. (9).

experimental data to Eq. (9), for each resonance S0 and

T1r(1H) values have been obtained (see Table 2).
The ratio (R) of moles of reacted DVS and moles of
polysaccharide rings can be calculated from the ratio
between a half of the area of the resonance due to methylene
carbons of type A, S0(A), and the area of anomeric carbons
C1 and C1(2p), S0(C1):
R

S0 A=2
S0 C1

10

The value of R for the case of sample H (R(H))was

calculated to be 0.28 ^ 0.03.
However since T1r(1H) values of the resonances,
respectively, due to C1 and A are not strictly the same
within the experimental errors, see Table 2, MAS-SPE
spectra were also performed.
In the case reported here, a reasonable RD value (40 s)
between successive scans allowed all carbon atoms to fully
relax back to equilibrium.
In Fig. 7 the 13C MAS-SPE spectra of sample L (a) and H
(b) are shown; note that the resonance A due to methylene
carbons has been evidenced.
Chemical shifts, line widths and areas obtained from
experimental spectra have been used as an input in a
program performing a full spectral deconvolution. The

deconvoluted spectra of sample L and H are shown in Fig.

7(c) and (d).
Using the areas obtained from the deconvoluted spectra,
from Eq. (10) R(L) and R(H) were evluated to be,
respectively, equal to 0.04 ^ 0.01 and 0.33 ^ 0.04 moles
of reacted DVS per mol of polysaccharide ring.
The detailed procedure for the evaluation of crosslnking
degree by means of 13C solid state NMR is reported in a
previuos paper [37].
From the value of R(L) it has been calculated that in
1 cm3 of crosslinked polymer sample L there were, on the
average, respectively, 0.0208 g of DVS and 0.9792 g of
polysaccharide fraction (assuming density of the dry
polymer equal to 1 g/cm3). It follows that there were
4.40 1023 moles of polysaccharide rings per cm3 of
crosslinked dry polymer and, on the average,
0.04(4.40 1023) moles of crosslinks per cm3 of dry
polymer. By using these results we obtained for the case of
sample L a value for rx equal to 3.50 1024 ^ 0.9
1024 moles/cm3 of crosslinked dry polymer.
Similarly, in the case of sample H, in 1 cm3 of crosslinked
polymer there were, on the average, respectively, 0.149 g of
DVS and 0.851 g of polysaccharide fraction (assuming again a
density of the dry polymer equal to 1 g/cm3). It follows that
there were 3.83 1023 moles of polysaccharide rings per cm3
of crosslinked dry polymer and, on the average,
0.33(3.83 1023) moles of crosslinks per cm3 of dry polymer.
By using these results we obtained for the case of sample H a
value for rx equal to 25 1024 ^ 3 1024 moles/cm3 of
crosslinked dry polymer.
4.4. Comparison of the three techniques
A substantial agreement among the results obtained for

Table 2
S0 and T1r(1H) values and the standard deviations for sample H obtained
fitting the experimental data to Eq. (9)
C atoms

S0

T1r(1H) (ms)

C1, C1(2p)
C4
C6
A

12.0 ^ 1
12.0 ^ 1
6.6 ^ 0.5
6.5 ^ 0.5

6.0 ^ 0.3
5.5 ^ 0.4
5.0 ^ 0.3
4.6 ^ 0.3

Fig. 7. 13C MAS-SPE experimental spectra of sample L (a) and H (b), on the
right the deconvoluted spectra (c) and (d) are shown.

F. Lenzi et al. / Polymer 44 (2003) 15771588

1587

Table 3
Values of rx with standard deviations as evaluated through free swelling, NMR and uniaxial compression experiments on samples L and H

Sample L
Sample H

Swelling

NMR

Uniaxial compression

3.50 1024 ^ 0.25 1024 mol/cm3

10.60 1024 ^ 2.5 1024 mol/cm3

3.50 1024 ^ 0.9 1024 mol/cm3

25 1024 ^ 3 1024 mol/cm3

2.4 1024 ^ 0.3 1024 mol/cm3

rx through the three different techniques was observed, as

summarized in Table 3.
The different approaches adopted provide different
results due to peculiar assumptions proper of each procedure
of evaluation as mentioned in the course of reported
calculations. However, the agreement is satisfactory and
some observations can be made. In fact, since NMR
technique gives results quite close to those obtained from
swelling and compression, it can be deduced that the
contribution of physical crosslinks is negligible, at least
right after occurrence of crosslinking reaction. This is likely
due to the fact that the low polymer concentration at which
crosslinking reaction was performed induced a rather small
amount of entanglements.
As reported previously, during desiccation step following crosslinking reaction, interactions slowly develop
promoting an increase of the effective physical crosslinking
(entanglements interactions) and, consistently, a marked
decrease of sorption capacity (ageing effect).

5. Conclusions
A comparison of a NMR based approach, aimed to the
determination of the degree of crosslinking of polymer
networks, with traditional experimental procedures has been
presented. It evidenced how the different experimental
approaches can be used sinergistically to gather detailed
information on the network structure. In fact, the methods
are sensitive to different chemical and physical structural
parameters of the macromolecular system.
The estimation of degree of crosslinking of a cellulose
based network as performed by swelling, compression and
NMR measurements, have been found to agree in the limits
of the experimental accuracy of the methods, although, in
the case of samples characterized by a higher degree of
crosslinking, NMR measurements lead to a slight overestimation, as compared to swelling experiments. This
conclusion is limited to the case of investigated degree of
crosslinking, which are those of interest for the application
of this material in the field of superabsorbing systems.
These results points to a substantial correspondence, for
the cellulose based system considered and for the adopted
crosslinking conditions, between chemically effective and
elastically effective crosslinks, at least in the limits of the
investigated chemical composition and of simplified
theoretical models adopted to interpret data.
Aging of dry samples has been found to promote a

decrease of swelling capacity, which is likely due to the

development of intermolecular interactions. Evidence of
this effect has also been found in the results of compression
experiments performed on swollen gel.

References
[1] Masuda F. Superabsorbent Polymers, Japan Polymer Society,
Kyoritsu Shuppann; 1987.
[2] Chambers DR, Fowler HH, Fujiura Y, Masuda F. Patent US 5,145,
906; 1992.
[3] Anbergen U, Opperman W. Polymer 1990;31:1854.
[4] Esposito F, Del Nobile MA, Mensitieri G, Nicolais L. J Appl Polym
Sci 1996;60:2403.
[5] Flory PJ. Principles of polymer chemistry. Ithaca, NY: Cornell
University Press; 1953.
[6] Brannon-Peppas L, Peppas NA. The equilibrium swelling behavior of
porous and non-porous hydrogels. In: Brannon-Peppas L, Harland RS,
editors. Absorbent polymer technology. Amsterdam: Elsevier; 1990.
p. 67 102.
[7] Harsh DC, Gehrke SH. Characterization of ionic water absorbent
polymers: determination of ionic content and effective crosslink
density. In: Brannon-Peppas L, Harland RS, editors. Absorbent
polymer technology. Amsterdam: Elsevier; 1990. p. 103 24.
[8] Baker PJ, Blanch HW, Prausnitz JM. Polymer 1995;36(5):10619.
[9] Ricka J, Tanaka T. Macromolecules 1984;17:2916 21.
[10] Konak C, Bansil R. Polymer 1989;30:677 80.
[11] Hooper HH, Baker PJ, Blanch HW, Prausnitz JM. Macromolecules
1990;23:1096 104.
[12] Flory PJ, Erman B. Macromolecules 1982;15:8006.
[13] Flory PJ. J Chem Phys 1942;10:5161.
[14] Treloar LRG. The physics of rubber elasticity, 3rd ed. Oxford, UK:
Oxford University Press; 1973.
[15] Huggins ML. Ann NY Acad Sci 1942;43:132.
[16] Erman B. Mechanical behavior of swollen networks. In: Buchholz FL,
Peppas NA, editors. Superabsorbent polymers. ACS Symposium
Series 573, Washington, DC: ACS; 1994. p. 5063. Chapter 4.
[17] Hatfield GE, Maciel GE. Macromolecules 1987;20:60815.
[18] Fyfe CA, Rudin A, Tchir WJ. Macromolecules 1980;13:1320 2.
[19] Garroway AN, Ritchey WM, Monix WB. Macromolecules 1982;15:
1051 63.
[20] Egger N, Schmidt-Rohr K, Bluemich B, Domke WD, Stapp B. J Appl
Polym Sci 1992;44:28995.
[21] Zhu XX, Banana K, Liu HY, Krause M, Yang M. Macromolecules
1999;32:277 81.
[22] Hietala S, Maunu SL, Sundholm F. Macromolecules 1999;32:
788 91.
[23] Law RV, Sherrington C, Snape CE, Ando I, Kurosu H. Macromolecules 1996;29:628493.
[24] Law RV, Sherrington C, Snape CE. Macromolcules 1997;30:
2868 75.
[25] Ebdon JR, Hunt BJ, ORourke WTS, Parkin J. Brit Polym J 1988;20:
32734.
[26] Chuang IS, Maciel GE. Macromolecules 1992;25:320426.
[27] Chuang IS, Maciel GE. Polymer 1994;35:1621 8.

1588

F. Lenzi et al. / Polymer 44 (2003) 15771588

[28] English AD, Chase DB, Spinelli HJ. Macromolecules 1983;16:

14227.
[29] Bauer DR, Dickie RA, Koenig JL. J Polym Sci, Polym Phys Ed 1984;
2:200920.
[30] Axelson DE, Nyhus AK. J Polym Sci, Part B: Polym Phys 1999;37:
130728.
[31] De Angelis A, Capitani D, Crescenzi V. Macromolecules 1998;5:
1595601.
[32] Yu G, Morin G, Nobes GAR, Marchessault RH. Macromolecules
1999;32:518 20.
[33] Yalpani M, Marchessault RH, Morin FG, Monasterios CJ. Macromolecules 1991;24:60469.
[34] Lindberg JJ, Hortling B. Adv Polym Sci 1985;66:122.
[35] Bray JC, Merrill EW. J Appl Polym Sci 1973;17:378196.
[36] Aklonis JJ, MacKnight WJ. Introduction to polymer viscoelasticity,
2nd ed. New York: Wiley-Interscience Publishers; 1983.
[37] Capitani D, Del Nobile MA, Mensitieri G, Sannino A, Segre AL.
Macromolecules 2000;33:4307.
[38] Capitani D, Porro F, Segre AL. Carbohydr Polym 2000;42(3):283 6.

[39] Anbergen U, Oppermann W. Polymer 1990;31:18548.

[40] Esposito F, Del Nobile MA, Mensitieri G, Nicolais L. J Appl Polym
Sci 1996;60:2403 7.
[41] Mehring M. Principles of high resolution NMR in solids. Berlin:
Springer; 1983.
[42] McBrierty V, Packer KJ. NMR in solid polymers. Cambridge, UK:
Cambridge University Press; 1993.
[43] Bain, AD Glinfit. Hamilton, Ontario L8S 4K1, Canada: Department of
Chemistry, McMaster University, 1989.
[44] Flory PJ. Discuss Faraday Soc 1970;49:7.
[45] Koningsveld R, Kleintjens LA, Schoffeleers HM. Pure Appl Chem
1974;39:1.
[46] Peppas NA, Lucht LM. Chem Engng Commun 1984;30:291.
[47] Voelkel R. In: Mathias LJ, editor. Solid state NMR of polymers. New
York: Plenum Press; 1988. p. 233 44. Chapter 13.
[48] Harris RK. Multidimensional magnetic resonance in liquids and
solidschemical application. In: Granger P, Harris RK, editors.
Dordrecht, The Netherlands: Kluwer Academic Publishers.